Between 1980 and 1985, 239 patients underwent mitral valve replacement with a Saint Jude Medical (130) or Carpentier-Edwards (109) prosthesis alone or associated with correction of tricuspid regurgitation. Age, NYHA grade and pulmonary artery pressures were comparable in the two groups. Two hundred and seventeen of the 221 survivors were contacted (98.6% follow-up): the long-term prosthesis-related mortality was 35% in the Saint Jude Medical group and 22.5% in the Carpentier-Edwards group. The global survival at 9 years was 82% in the Saint Jude Medical group but only 63% in the Carpentier-Edwards group. The actuarial thromboembolic-free rate at 9 years was 96% in the Saint Jude Medical and 86% in the Carpentier-Edwards group (not significant). Two Saint Jude Medical and one Carpentier-Edwards bioprosthesis thrombosed--a linear rate of 0.2% PY and 0.1% PY respectively. Furthermore, there was no significant difference in the actuarial rate of haemorrhagic complications between the two groups. Endocarditis and valve degeneration were more common in the Carpentier-Edwards group (1.3% and 1.6% PY respectively). Over the 9 year period, the thromboembolic complications of the Saint Jude Medical and Carpentier-Edwards prostheses were similar in actuarial and linear terms. With respect to long-term mortality and morbidity, the results with the Saint Jude Medical valve seem to be superior to those of the Carpentier-Edwards bioprosthesis.